Ontology-based Queriesover Cancer Data
Alejandra González-Beltrán1,2,Ben Tagger1, Anthony Finkelstein1
1Department of Computer Science 2Computational and Systems MedicineUniversity College London
London, United Kingdom
Semantic Web Applications and Tools for Life Sciences(SWAT4LS)
Berlin, Germany10th December 2010
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Overview• Motivation: queries over cancer data• Background: caGrid infrastructure
• caGrid query language (CQL)
• Objective: ontology-based queries over the caGridinfrastructure
• Approach:• OWL representation of caGrid models• Query rewriting & translation
• Implementation & performance evaluation• Conclusions
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Overview• Motivation: queries over cancer data• Background: caGrid infrastructure
• caGrid query language (CQL)
• Objective: ontology-based queries over the caGridinfrastructure
• Approach:• OWL representation of caGrid models• Query rewriting & translation
• Implementation & performance evaluation• Conclusions
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Overview• Motivation: queries over cancer data• Background: caGrid infrastructure
• caGrid query language (CQL)
• Objective: ontology-based queries over the caGridinfrastructure
• Approach:• OWL representation of caGrid models• Query rewriting & translation
• Implementation & performance evaluation• Conclusions
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Overview• Motivation: queries over cancer data• Background: caGrid infrastructure
• caGrid query language (CQL)
• Objective: ontology-based queries over the caGridinfrastructure
• Approach:• OWL representation of caGrid models• Query rewriting & translation
• Implementation & performance evaluation• Conclusions
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Overview• Motivation: queries over cancer data• Background: caGrid infrastructure
• caGrid query language (CQL)
• Objective: ontology-based queries over the caGridinfrastructure
• Approach:• OWL representation of caGrid models• Query rewriting & translation
• Implementation & performance evaluation• Conclusions
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
MotivationCancer researcher interested in the changes in chromosome 17(associated with prostate, bladder, breast cancers) wants to
find single nucleotide polymorphisms (SNPs)associated with chromosome 17
Software infrastructures to manage and analyse cancer datafrom heterogeneous data sources
• UK National Cancer Research Institute (NCRI)Informatics Initiative: ONcology Information eXchange(ONIX)
• US National Cancer Institute (NCI) caBIG® programme:caGrid infrastructure
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
MotivationCancer researcher interested in the changes in chromosome 17(associated with prostate, bladder, breast cancers) wants to
find single nucleotide polymorphisms (SNPs)associated with chromosome 17
Software infrastructures to manage and analyse cancer datafrom heterogeneous data sources
• UK National Cancer Research Institute (NCRI)Informatics Initiative: ONcology Information eXchange(ONIX)
• US National Cancer Institute (NCI) caBIG® programme:caGrid infrastructure
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
caGrid InfrastructureService-oriented, model-driven infrastructure.
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
caGrid Query Language (CQL)N
atur
e P
rece
ding
s : d
oi:1
0.10
38/n
pre.
2010
.540
6.1
: Pos
ted
16 D
ec 2
010
caGrid Query Language (CQL)N
atur
e P
rece
ding
s : d
oi:1
0.10
38/n
pre.
2010
.540
6.1
: Pos
ted
16 D
ec 2
010
caGrid Query Language (CQL)N
atur
e P
rece
ding
s : d
oi:1
0.10
38/n
pre.
2010
.540
6.1
: Pos
ted
16 D
ec 2
010
caGrid Query Language (CQL)N
atur
e P
rece
ding
s : d
oi:1
0.10
38/n
pre.
2010
.540
6.1
: Pos
ted
16 D
ec 2
010
caGrid Query Language (CQL)N
atur
e P
rece
ding
s : d
oi:1
0.10
38/n
pre.
2010
.540
6.1
: Pos
ted
16 D
ec 2
010
caGrid Query Language (CQL)N
atur
e P
rece
ding
s : d
oi:1
0.10
38/n
pre.
2010
.540
6.1
: Pos
ted
16 D
ec 2
010
caGrid Query Language (CQL)
Navigational (path-finding) query language over the structureof caGrid data resources
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Motivation — revisitedcaBIO data service (cancer Bioinformatics InfrastructureObjects) — biomedical data from a variety of curated datasources
<ns1:CQLQuery xmlns:ns1="http://CQL.caBIG/1/gov.nih.nci.cagrid.CQLQuery"><ns1:Target name="gov.nih.nci.cabio.domain.SNP"><ns1:Association name="gov.nih.nci.cabio.domain.SNPPhysicalLocation"roleName="physicalLocationCollection"><ns1:Association name="gov.nih.nci.cabio.domain.Chromosome" roleName="chromosome">
<ns1:Attribute name="number" predicate="EQUAL_TO" value="17"/></ns1:Association>
</ns1:Association></ns1:Target>
</ns1:CQLQuery>
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Motivation — revisitedcaBIO data service (cancer Bioinformatics InfrastructureObjects) — biomedical data from a variety of curated datasources
<ns1:CQLQuery xmlns:ns1="http://CQL.caBIG/1/gov.nih.nci.cagrid.CQLQuery"><ns1:Target name="gov.nih.nci.cabio.domain.SNP"><ns1:Association name="gov.nih.nci.cabio.domain.SNPPhysicalLocation"roleName="physicalLocationCollection"><ns1:Association name="gov.nih.nci.cabio.domain.Chromosome" roleName="chromosome">
<ns1:Attribute name="number" predicate="EQUAL_TO" value="17"/></ns1:Association>
</ns1:Association></ns1:Target>
</ns1:CQLQuery>
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Motivation — revisitedcaBIO data service (cancer Bioinformatics InfrastructureObjects) — biomedical data from a variety of curated datasources
<ns1:CQLQuery xmlns:ns1="http://CQL.caBIG/1/gov.nih.nci.cagrid.CQLQuery"><ns1:Target name="gov.nih.nci.cabio.domain.SNP"><ns1:Association name="gov.nih.nci.cabio.domain.SNPPhysicalLocation"roleName="physicalLocationCollection"><ns1:Association name="gov.nih.nci.cabio.domain.Chromosome" roleName="chromosome">
<ns1:Attribute name="number" predicate="EQUAL_TO" value="17"/></ns1:Association>
</ns1:Association></ns1:Target>
</ns1:CQLQuery>
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Objective: ontology-based queries
Semantic layer on top of caGrid structural layer
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Objective: ontology-based queries
Semantic layer on top of caGrid structural layer
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Objective: ontology-based queries
Semantic layer on top of caGrid structural layer
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Overview• Motivation: queries over cancer data• Background: caGrid infrastructure
• caGrid query language (CQL)
• Objective: ontology-based queries over the caGridinfrastructure
• Approach:• OWL representation of caGrid models• Query rewriting & translation
• Implementation & performance evaluation• Conclusions
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
OWL representation of caGrid models
UML class diagrams
c:Chromosome ⊑ u:UMLClassc:number ⊑ u:UMLAttributec:number ⊑ ∃u:hasValue.xsd:string
c:locationCollection ⊑ u:hasAssociation
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
OWL representation of caGrid models
UML class diagrams
c:PhysicalLocation ⊑ c:Locationc:Chromosome ⊑ ∃c:locationCollection.c:Locationc:Chromosome ⊑ ∃u:hasAttribute.c:number
c:PhysicalLocation ⊑ ∃c:chromosome.c:Chromosome
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
OWL representation of caGrid models
Semantic annotations
c:SNP ⊑ n:Single_Nucleotide_Polymorphismc:Chromosome ⊑ n:Chromosome
c:number ⊑ n:Name
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
OWL representation of caGrid models
Module extraction from NCIt• Each caGrid information model refers to a subset Σ of theNCIt vocabulary — relevant terms and relationships
• NCIt module for each data model: Logic-based moduleextraction
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Query rewriting and translation
ontology-based query →CQL
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Query rewriting and translation
Parsingn:Single_Nucleotide_Polymorphism and hasAssociation some(n:Chromosome and hasAttribute some (n:Name and hasValue value"17"))
UML Extractionc:SNP and hasAssociation some (c:Chromosome and hasAttribute some(c:number and hasValue value "17"))
Data Values Extractionc:SNP and hasAssociation some (c:Chromosome and hasAttribute some(c:number))
Semantic ValidationQuery satisfiable in the ontology?
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Query rewriting and translation
Parsingn:Single_Nucleotide_Polymorphism and hasAssociation some(n:Chromosome and hasAttribute some (n:Name and hasValue value"17"))
UML Extractionc:SNP and hasAssociation some (c:Chromosome and hasAttribute some(c:number and hasValue value "17"))
Data Values Extractionc:SNP and hasAssociation some (c:Chromosome and hasAttribute some(c:number))
Semantic ValidationQuery satisfiable in the ontology?
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Query rewriting and translation
Parsingn:Single_Nucleotide_Polymorphism and hasAssociation some(n:Chromosome and hasAttribute some (n:Name and hasValue value"17"))
UML Extractionc:SNP and hasAssociation some (c:Chromosome and hasAttribute some(c:number and hasValue value "17"))
Data Values Extractionc:SNP and hasAssociation some (c:Chromosome and hasAttribute some(c:number))
Semantic ValidationQuery satisfiable in the ontology?
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Query rewriting and translation
Parsingn:Single_Nucleotide_Polymorphism and hasAssociation some(n:Chromosome and hasAttribute some (n:Name and hasValue value"17"))
UML Extractionc:SNP and hasAssociation some (c:Chromosome and hasAttribute some(c:number and hasValue value "17"))
Data Values Extractionc:SNP and hasAssociation some (c:Chromosome and hasAttribute some(c:number))
Semantic ValidationQuery satisfiable in the ontology?
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Query rewriting and translation
Properties Path Finderc:SNP and c:physicalLocationCollection some c:SNPPhysicalLocation andc:chromosome some (c:Chromosome and hasAttribute some (c:number))
Data Values Additionc:SNP and c:physicalLocationCollection some c:SNPPhysicalLocation andc:chromosome some (c:Chromosome and hasAttribute some (c:numberand hasValue value "17"))
OWL Expression to MCC Translation⊎ { s 8 s ← SNP, r ← s.physicalLocationCollection, r ←
SNPPhysicalLocation, c ← r.chromosome, c ← Chromosome,c.number=17 }
MCC to CQL TranslationCQL Query
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Query rewriting and translation
Properties Path Finderc:SNP and c:physicalLocationCollection some c:SNPPhysicalLocation andc:chromosome some (c:Chromosome and hasAttribute some (c:number))
Data Values Additionc:SNP and c:physicalLocationCollection some c:SNPPhysicalLocation andc:chromosome some (c:Chromosome and hasAttribute some (c:numberand hasValue value "17"))
OWL Expression to MCC Translation⊎ { s 8 s ← SNP, r ← s.physicalLocationCollection, r ←
SNPPhysicalLocation, c ← r.chromosome, c ← Chromosome,c.number=17 }
MCC to CQL TranslationCQL Query
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Query rewriting and translation
Properties Path Finderc:SNP and c:physicalLocationCollection some c:SNPPhysicalLocation andc:chromosome some (c:Chromosome and hasAttribute some (c:number))
Data Values Additionc:SNP and c:physicalLocationCollection some c:SNPPhysicalLocation andc:chromosome some (c:Chromosome and hasAttribute some (c:numberand hasValue value "17"))
OWL Expression to MCC Translation⊎ { s 8 s ← SNP, r ← s.physicalLocationCollection, r ←
SNPPhysicalLocation, c ← r.chromosome, c ← Chromosome,c.number=17 }
MCC to CQL TranslationCQL Query
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Query rewriting and translation
Properties Path Finderc:SNP and c:physicalLocationCollection some c:SNPPhysicalLocation andc:chromosome some (c:Chromosome and hasAttribute some (c:number))
Data Values Additionc:SNP and c:physicalLocationCollection some c:SNPPhysicalLocation andc:chromosome some (c:Chromosome and hasAttribute some (c:numberand hasValue value "17"))
OWL Expression to MCC Translation⊎ { s 8 s ← SNP, r ← s.physicalLocationCollection, r ←
SNPPhysicalLocation, c ← r.chromosome, c ← Chromosome,c.number=17 }
MCC to CQL TranslationCQL Query
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Implementation & performance evaluation• Two modules: OWL generator (exposed as a caGridanalytical service) & query rewriting/translation
• Java, caGrid 1.3, OWLAPI 3.1, Pellet 2.2.2, HermiT 1.3.0
• Performance
• Analysis of generated OWL ontologies (caGrid models)— path metrics
• Ontology generation, module extraction & classification• Query rewriting/translation
• Three groups of caGrid models
• caDSR — registered in caDSR• caGrid — registered in caGrid index service• InfoModels — models supported by deployed services
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Implementation & performance evaluation• Two modules: OWL generator (exposed as a caGridanalytical service) & query rewriting/translation
• Java, caGrid 1.3, OWLAPI 3.1, Pellet 2.2.2, HermiT 1.3.0
• Performance• Analysis of generated OWL ontologies (caGrid models)— path metrics
• Ontology generation, module extraction & classification• Query rewriting/translation
• Three groups of caGrid models
• caDSR — registered in caDSR• caGrid — registered in caGrid index service• InfoModels — models supported by deployed services
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Implementation & performance evaluation• Two modules: OWL generator (exposed as a caGridanalytical service) & query rewriting/translation
• Java, caGrid 1.3, OWLAPI 3.1, Pellet 2.2.2, HermiT 1.3.0
• Performance• Analysis of generated OWL ontologies (caGrid models)— path metrics
• Ontology generation, module extraction & classification• Query rewriting/translation
• Three groups of caGrid models• caDSR — registered in caDSR• caGrid — registered in caGrid index service• InfoModels — models supported by deployed services
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Analysis of OWL representation
Path Metrics
• Longest path: up to 36 nodes; for 75 % of the projects ineach category their length is less than 17 or 18
• Average path length: median between 4 and 7 nodes; for75 % of the InfoModels it is less than 8
• Average paths per journey: median ∼ 2 paths per journey;for 75 % of the projects (3 categories), less than 2.5
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Analysis of OWL representation
Path Metrics
• Longest path: up to 36 nodes; for 75 % of the projects ineach category their length is less than 17 or 18
• Average path length: median between 4 and 7 nodes; for75 % of the InfoModels it is less than 8
• Average paths per journey: median ∼ 2 paths per journey;for 75 % of the projects (3 categories), less than 2.5
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Analysis of OWL representation
Path Metrics
• Longest path: up to 36 nodes; for 75 % of the projects ineach category their length is less than 17 or 18
• Average path length: median between 4 and 7 nodes; for75 % of the InfoModels it is less than 8
• Average paths per journey: median ∼ 2 paths per journey;for 75 % of the projects (3 categories), less than 2.5
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Analysis of OWL representation
Path Metrics
• Longest path: up to 36 nodes; for 75 % of the projects ineach category their length is less than 17 or 18
• Average path length: median between 4 and 7 nodes; for75 % of the InfoModels it is less than 8
• Average paths per journey: median ∼ 2 paths per journey;for 75 % of the projects (3 categories), less than 2.5
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Ontologies, modules & classification
Generation and inference times
• 75 % of NCIt modules, extraction takes less than 2seconds & even less time for ontology generation
• median inference time (Pellet & HermiT reasoners): lessthan 100 ms
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Ontologies, modules & classification
Generation and inference times
• 75 % of NCIt modules, extraction takes less than 2seconds & even less time for ontology generation
• median inference time (Pellet & HermiT reasoners): lessthan 100 ms
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Query rewriting/translation evaluation
Query rewriting — path lengths 1 and 2, and mean values
• Stages: (1) parsing, (2) UML extraction, (3) validation,(4) path finding, (5) MCC conversion and (6) CQLconversion
• Path length: affects path-finding stage, rest of stagesremain largely unaffected.
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Work in progress...• Explore OWL2EL reasoners — improve path finding stage• Building a query suite• GUI development
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Work in progress...• Explore OWL2EL reasoners — improve path finding stage• Building a query suite• GUI development
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Conclusions• Ontology-based queries over caGrid: design &implementation
• General approach for service-oriented, model-driveninfrastructure — only last step of query rewriting(MCC2CQL) depends on caGrid
• Generation of OWL2 ontologies from annotated UMLmodels (ISO11179 standard)
• Analysis of generated ontologies — path metric• caGrid analytical service for the OWL generator• Analysis of CQL• Query rewriting/translation procedure — OWL classexpressions →MCC →CQL
• Performance evaluation — OWL generation, moduleextraction, classification
• Assessment of query rewriting/translation procedure andits viability
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Conclusions• Ontology-based queries over caGrid: design &implementation
• General approach for service-oriented, model-driveninfrastructure — only last step of query rewriting(MCC2CQL) depends on caGrid
• Generation of OWL2 ontologies from annotated UMLmodels (ISO11179 standard)
• Analysis of generated ontologies — path metric• caGrid analytical service for the OWL generator• Analysis of CQL• Query rewriting/translation procedure — OWL classexpressions →MCC →CQL
• Performance evaluation — OWL generation, moduleextraction, classification
• Assessment of query rewriting/translation procedure andits viability
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Conclusions• Ontology-based queries over caGrid: design &implementation
• General approach for service-oriented, model-driveninfrastructure — only last step of query rewriting(MCC2CQL) depends on caGrid
• Generation of OWL2 ontologies from annotated UMLmodels (ISO11179 standard)
• Analysis of generated ontologies — path metric• caGrid analytical service for the OWL generator• Analysis of CQL• Query rewriting/translation procedure — OWL classexpressions →MCC →CQL
• Performance evaluation — OWL generation, moduleextraction, classification
• Assessment of query rewriting/translation procedure andits viability
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Conclusions• Ontology-based queries over caGrid: design &implementation
• General approach for service-oriented, model-driveninfrastructure — only last step of query rewriting(MCC2CQL) depends on caGrid
• Generation of OWL2 ontologies from annotated UMLmodels (ISO11179 standard)
• Analysis of generated ontologies — path metric• caGrid analytical service for the OWL generator• Analysis of CQL• Query rewriting/translation procedure — OWL classexpressions →MCC →CQL
• Performance evaluation — OWL generation, moduleextraction, classification
• Assessment of query rewriting/translation procedure andits viability
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Conclusions• Ontology-based queries over caGrid: design &implementation
• General approach for service-oriented, model-driveninfrastructure — only last step of query rewriting(MCC2CQL) depends on caGrid
• Generation of OWL2 ontologies from annotated UMLmodels (ISO11179 standard)
• Analysis of generated ontologies — path metric• caGrid analytical service for the OWL generator• Analysis of CQL• Query rewriting/translation procedure — OWL classexpressions →MCC →CQL
• Performance evaluation — OWL generation, moduleextraction, classification
• Assessment of query rewriting/translation procedure andits viability
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Conclusions• Ontology-based queries over caGrid: design &implementation
• General approach for service-oriented, model-driveninfrastructure — only last step of query rewriting(MCC2CQL) depends on caGrid
• Generation of OWL2 ontologies from annotated UMLmodels (ISO11179 standard)
• Analysis of generated ontologies — path metric• caGrid analytical service for the OWL generator• Analysis of CQL• Query rewriting/translation procedure — OWL classexpressions →MCC →CQL
• Performance evaluation — OWL generation, moduleextraction, classification
• Assessment of query rewriting/translation procedure andits viability
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Conclusions• Ontology-based queries over caGrid: design &implementation
• General approach for service-oriented, model-driveninfrastructure — only last step of query rewriting(MCC2CQL) depends on caGrid
• Generation of OWL2 ontologies from annotated UMLmodels (ISO11179 standard)
• Analysis of generated ontologies — path metric• caGrid analytical service for the OWL generator• Analysis of CQL• Query rewriting/translation procedure — OWL classexpressions →MCC →CQL
• Performance evaluation — OWL generation, moduleextraction, classification
• Assessment of query rewriting/translation procedure andits viability
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Conclusions• Ontology-based queries over caGrid: design &implementation
• General approach for service-oriented, model-driveninfrastructure — only last step of query rewriting(MCC2CQL) depends on caGrid
• Generation of OWL2 ontologies from annotated UMLmodels (ISO11179 standard)
• Analysis of generated ontologies — path metric• caGrid analytical service for the OWL generator• Analysis of CQL• Query rewriting/translation procedure — OWL classexpressions →MCC →CQL
• Performance evaluation — OWL generation, moduleextraction, classification
• Assessment of query rewriting/translation procedure andits viability
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Conclusions• Ontology-based queries over caGrid: design &implementation
• General approach for service-oriented, model-driveninfrastructure — only last step of query rewriting(MCC2CQL) depends on caGrid
• Generation of OWL2 ontologies from annotated UMLmodels (ISO11179 standard)
• Analysis of generated ontologies — path metric• caGrid analytical service for the OWL generator• Analysis of CQL• Query rewriting/translation procedure — OWL classexpressions →MCC →CQL
• Performance evaluation — OWL generation, moduleextraction, classification
• Assessment of query rewriting/translation procedure andits viability
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Acknowledgments• National Cancer Research Institute (NCRI) InformaticsInitiative, UK
• Konrad Rokicki, SAIC/NCI CBIIT
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0
Thank you!
Questions?
Nat
ure
Pre
cedi
ngs
: doi
:10.
1038
/npr
e.20
10.5
406.
1 : P
oste
d 16
Dec
201
0